Measurement error in physical activity assessment has made it difficult to answer important questions about the prevalence of physical activity and associations with various health-related outcomes. The objective in the present application is to develop and evaluate statistical procedures to model, quantify and adjust for measurement error in a commonly used and accepted physical activity recall instrument (24 hour physical activity recall 24hPAR). To develop appropriate statistical models, we will conduct a Physical Activity Measurement Survey (PAMS) to obtain recall and objective physical activity data from a representative sample of 1200 adults (19- 70 yrs) who reside in rural and urban environments in Iowa. Participants will complete two days of physical activity monitoring with the BodyMedia Sensewear Pro III (SP3), a multi-channel pattern recognition device that provides accurate estimates of PA and energy expenditure. After each monitoring day, participants will complete a telephone-administered 24hPAR assessment. After replicate measures of the SP3 and 24hPAR are obtained, self-reported physical activity propensity data will be obtained to provide auxiliary information for estimating models and distributions of usual physical activity. The sequential series of Specific Aims will address unique questions about measurement error in physical activity and lead to the development of procedures and methodologies to facilitate the application of these methods in future research. In Aim 1, we will develop a self- administered physical activity propensity questionnaire (PAPQ) and a 24 hr PA recall (24hPAR) telephone interview that can be administered in a large-scale survey setting. In Aim 2, measurement error models will be estimated for the recall and reference measures in order to estimate the bias and random measurement error structure of measurements. A unique aspect of the proposed modeling is that we will utilize new propensity-based approaches to address the fact that many adults in the population report no physical activity. In Aim 3, the 24PAR will be calibrated against the temporally matched SP3 data so that the measurements essentially behave as if they had been collected using a reference instrument. In Aim 4, the measurement error model and calibration procedures will be used to estimate usual daily physical activity of individuals in subpopulations. The approach in this research is innovative, because it utilizes state of the art monitors and will lead to the development of new statistical techniques to model and correct physical activity measurement error. The proposed research is significant, because it will directly address a complex and long-standing measurement problem in the physical activity field (i.e. obtaining accurate indicators of usual physical activity). This information will help to improve physical activity epidemiology research and facilitate the development of more effective public health surveillance research. The resulting physical activity measurement model from this study will also facilitate future research aimed at jointly modeling error in energy intake and energy expenditure. The project is guided by a strong research team with expertise in all necessary facets of the study (physical activity assessment, survey design and administration, and measurement error modeling).